Videoimage analysis and objects identification

Development of videoimage algorithms analysis and objects recognition technologies, increasing "intellectuality" of video surveillance systems considerably expands their sphere of application. For example, by means of video surveillance it is possible to find out automatically failures in productions (for example, a jam on a conveyor line), to keep account finished goods and materials, to analyze quality. Video surveillance "intellectual" systems in stadiums and concert halls allow revealing people with inadequate behavior. The video analysis of the transport streams in a mode of real time is used for traffic intensity definition and accidents prophylactics. The visual analysis of passing trains technical condition provides duly revealing malfunctions and preventive maintenance of failures. Analyzing videosystems can be used for automation of many processes.

The module of automobile numbers recognition

System is used automatic automobile numbers registration and recognition on check items at the enterprises, paid parking and garage complexes, posts of motor licensing and inspection department. Integration with the access-control system allows automating a check mode. The system allows conducting searching cars in stealing.

Functionalities:

• Automatic registration of automobile numbers and recognition;
• Saving numbers and videorecording vehicles traveling in a database with the date and time;
• Automatic comparison automobile number and available databases (for example, cars which are having the right of the admission on the territory of the company, or cars which are being stealing) and delivery of the corresponding message to the operator;
• Automation check mode and integration with devices of the access control;
• Search in a database by number, date, time;
• Reports formation by number, date, time.

Characteristics:

• Width of a control zone - from 1,5 up to 3,5 meters
• Depth of the control zone - 10 meters (for a check point: 2-4 meters)
• Quantity of frames (with 100% streaming congestion on all cameras):
  1 channel - 25 frame/second up to 150 km/h
  2 channels and more - 16 fr/s (total) up to 75 km/h
• Probability of recognition - not less than 90 %
• Admissible corners of videocamera installation:
  On a vertical - no more than 30°;
  Across - no more than 20°;
• The Roll of license plate ±15°;
• Light exposure - not less than 50 lux
• Distinguished numbers - 7 types (a background white and yellow)
• Requirement to a database - support of interface ADO
• Minimal configuration of the PC - Celeron 1700 256MB Windows XP

Capture and recognition of the face

The system of faces capture allows allocating only people faces, to choose the most distinct image from several variants and to keep them in a database. The system of the faces recognition identifies the person and automates images searching in databases.

Quality control and the account of made production

Technology of recognition used for automatic quality production control and the account of its quantity allows to accelerate process of manufacture and to prevent plunders. The account of production is carried out at each stage of manufacture, along the line on a warehouse and then to the buyer. At discrepancy of production quantity planned the system automatically gives out the message to the operator and the production manager. The account of production at each stage allows defining automatically on what them and during what time plunder is carried out. Videorecordings of each unit movement also are stored in archives at each stage. The monitoring system of quality and production quantity is developed under an individual task in conformity with specificity of each concrete manufacture.

The module of freight carriage recognition

The system uses for an automation registration and recognition of freight carriage numbers and cistern numbers at the rail junctions, transit exchanges, the approaches to the unloading points/the load points.
The demands to the source of video signal:
The camera permission is from 300 tvl by horizontal
The camera sensitivity is by F1.2, ES –1/50 from 0.003 lux.
Relations of the moving speed for a freight car with the install parameter of the electron camera lock:
Up to 20 km/h – 1/500 sec;
Up to 40 km/h – 1/1000 sec;
Up to 100 km/h – 1/2500 sec.
The system makes the high demands to the combination of parameters: maximum moving speed – the camera sensitivity – the artificial lightning, which are defined at the project phase.
The format of signal digitization – black-and-white 768*288 pixels.

The parameters of camera installing:

The wide of number reading zone – from 2,5 up to 4 meters.
An optimal distance from camera to railway road bed 4-9 m
Angle of incidence for a videocamera to the railway road bed - 90? (±15?).
Horizontal camera roll - ±10?.
Skewing of the camera is no more than 30?.

The characteristics of multichannel regime:

A bundling of the recognition cameras and the review cameras into the unit system.
The number of channels is limited by the computer performance. The possible combination of channels is 15+1 (15 review cameras + 1 recognition), 14+2, 12+4 and etc.
The quantity of permissible recognition channels for one computer (with 100% streaming workload at all channels):

• 1 channel – 12 fps up to 60 km/h
• 2 channel – 6 fps up to 30 km/h
• 4 channel – 3 fps up to 15 km/h

The characteristic of numbers recognition:

The recognition of freight carriage: the roofed freight carriage, the gondola carriage, the cisterns, the insulated carriage and other specialized freight carriage.
Colour number characteristics – dark at the clear background, the clear at the dark background.
A control of carriage number with the using of control number.

The user’s interface:

An authentication of system users and their rights.
The output at the screen a videoimage from the recognition camera in real time.
The displaying of last recognized number at the screen at any camera.
A fast access to the video archive by the event.
An editing of recognized numbers. The input of new number by the absence of the recognized with a clamping to the image with using of video recording.
A sound signaling of the disturbed and information events.
The output of carriage characteristics at the monitor.

The input of database:

The using of InterBase. SQL Server.
The possibility of centralized information keeping (several check points – one server).
A fixation of the information about carriages in database, passage through the check point. The latched data: name of check point, the number of gates at the check point, the date and time of passage, the recognition number, the corrected number, the operator’s name, the sign of passage (by sensors), an image from camera at the recognition/passage moment.
The conduct of black and white bills (the bill of detention, the bill to the pass through the check point). The recognized number controls in inside database on a base SQL-Server InterBase, and also activates the keeping procedure of outside data base, using ADO connection;
A fixation of system events (the system activation, the system switching-out, the rebooting, the blockage of sensors and so on).
The search in database by number, date, time.
A flexible drawing up of accounts by events, happened at the check point, permission/ disabling to the passage.

An additional possibilities:

The information keeping (the number and the video recording of passage) into the video archive with the direction of date and time no less than 15 days;
An automation of the check point regime (by the integration with the access control devices). On a base of information of inside database and outside enquiry, the passage is able or disable. The information displays at the monitor and happens the sound signaling for an operator. The program may manage the outside devices (barrier, light signal and so on), and also to gather the information about passages from sensors.